The invention is related to the field of direct-sequence spread spectrum communication systems such as those implementing the CDMA-2000, UMTS, IS-95 standards and similar cellular telephone systems which apply a pseudo-noise sequence for encoding and decoding data.
Spread spectrum communication systems are finding increased use in two-way aerial communication. Just as AM and FM systems use a sinusoidal signal to carry information, spread spectrum systems use a noise-like signal to carry information. In a transmitter, a stream of digital data is encoded with a pseudo-noise sequence (PNS) to spread the spectrum of the signal for transmitting the data through a media. At a receiver the data is recovered from the media and then decoded using the same PNS to de-spread the spectrum of the signal to reproduce the original digital data stream.
A PNS is a stream of bits with a pattern that is determinate, but which appears to be a random bit stream. A common apparatus for producing a PNS is a linear feedback shift register (LFSR). Two common types of LFSRs are Fibonacci LFSRs and Galois LFSRs. Both types include a closed loop circuit containing bit registers and modulo-2 adders through which bits are shifted through the loop. The adders have one input that is part of the loop and another input that is connected to another part of the loop to form multiple loops to randomize the bits as they are shifted through the loop.
The values of the PNS for any LFSR, repeat after a large number of bits and it is desirable to provide a PNS with the longest possible sequence without repeating using a limited amount of hardware. This is accomplished by choosing the configuration of the adders and registers of the LFSR and the initial values of the registers in a manner well known in the art. For a given number of registers m contained in the LFSR, the longest possible non-repeating portion of the PNS is equal in length to 2mxe2x88x921 bits.
In addition to using the same PNS, the transmitter and receiver must use values from the same position in the PNS for spreading and de-spreading respectively. In order to synchronize the transmitter and receiver to both use values at the same position in the PNS, an offset mask value is calculated and combined with the output values of the current position of the PNS (in the transmitter or receiver) to produce the values of a different shifted position in the PNS in a manner well known in the art.
Those skilled in the art are directed to the following citations. U.S. Pat. No. 5,878,076 to Siedenburg describes a direct sequence spread spectrum communication system. U.S. Pat. No. 5,754,603 to Thomas describes PNS synchronization. U.S. Pat. No. 5,926,070 to Barron describes offset mask generation. European patent application publication 0 660 541 by Ishida describes methods of synchronizing PNS positions of a transmitter and receiver. PCT patent application publication WO 99/45670 by Medlock describes masks for LFSRs.
FIG. 1 describes selected portions of a Galois LFSR with an offset mask. LFSR 100 includes a multitude of binary registers 101-108 connected in series in a loop circuit. The binary registers may be D-flip-flops or other know bit storage devices. Using register 102 as an example, each register 102 has a value input 110 connected to an output 111 of a previous register 101 and each register 102 has an output 112 connected to the value input 113 of a subsequent register 103.
LFSR 100 also includes one or more modulo-2 adders 115-117 connected in the loop circuit. Each adder is inserted between a different pair of sequential registers 101-108 of the register series. The selection of the pairs of registers between which adders are inserted, depends on the selection of a primitive binary polynomial. A primitive polynomial is similar in concept to a prime number. A primitive polynomial is a polynomial that can not be divided by any simpler polynomial. For the specific example LFSR shown in FIG. 1, the primitive binary polynomial is D8+D4+D3+D2+1. The D8 requires the LFSR to have 8 registers, and the D2, D3 and D4 terms require adders be inserted between the second to the last, third from the last, and fourth from the last pairs of registers as shown. Primitive polynomials, like prime numbers, are well known in the art.
The inserted adders 115-117 each have two inputs and one output and may be simply implemented as XOR gates. As an example, adder 115 has first input 120 connected to output 121 of previous register 104 and output 122 connected to value input 123 of subsequent register 105. Also, adder 115 has second input 124 connected between output 125 of last register 108 and input 126 of first register 101 of the register series. Clock signal line 130 is connected to a clock input of each register of the register series, and when a clock signal is transmitted through the clock signal line, each register begins to output the value being received at that time at the register""s value input. For example, clock signal line 130 is connected to clock input 131 of register 101.
Control lines 135 includes at least one initialization line 136 connected to each register 101-108 in order to initialize the values of the registers. For example, initialization line 136 is shown connected to initialization input 137 of register 108. The initialization line may write a memory value into the register so that any initial value can be written into any register as desired. In that case, the initial values of the registers are usually predetermined and stored in a memory. Alternatively, the control line may simply signal the register to assume some predetermined initial value that is built into the hardware of the particular register. If the registers are D-flip-flops the initialization line is connected to the set input of every register to be initialized to one and connected to the reset input of every register to be initialized to zero, and when the initialization line goes high, the values of the registers assume their respective initial values. Methods for selecting the initial values of the registers for a particular primitive polynomial are well known and further discussion is not required herein.
The Galois LFSR shown in FIG. 1 outputs bit values for the PNS at output 138. However, in order for a receiver to synchronize the position of the output values in the PNS with the position of output values for a transmitter using the same PNS (or vice versa), offset mask values must be combined with a previously output portion of the PNS.
Mask 140 is connected with output 138 of Galois LFSR 100 as shown in FIG. 1. The mask includes a series of registers 141-148 which respectively store the previous 8 values of the PNS output from the LFSR. The outputs of registers 142-148 are connected to the inputs of respective subsequent registers 141-147. For example, input 149 of register 146 is connected to output 150 of register 147, and output 151 of register 146 is connected to the input 152 of register 145.
The mask also includes a series of modulo-2 adders 161-167 with a first input of each subsequent adder 162-167 connected to an output of a previous respective adder 161-166 in the adder series. For example, input 153 of adder 165 is connected to output 154 of adder 164 and output 155 of adder 165 is connected to input 156 of adder 166. A multitude of mask switches 171-178 include a first mask switch 171 with an output 179 connected to a first input 180 of first adder 161 of the adder series. Also, subsequent mask switches 172-178 have outputs connected to respective second inputs of adders 161-167 in the adder series. The output of each register 141-148 is connected to the input of respective switches 171-178.
Mask value lines 191-198 of control lines 135 are connected respectively to switches 171-178, in order to set respective switches 171-178 in an open or closed position which controls whether the value of a respective register is provided through the respective switch to a respective input of a respective adder of adders 161-167. For example, output 151 of register 146 is connected to input 182 of switch 175 and output 183 of switch 175 is connected to input 184 of adder 165. Thus, when switch line 196 is set to 1, then the value of register 146 is modulo-2 added with the output value of output 154 of adder 164 and the result is output at output 155 to input 156 of adder 166. Otherwise, when switch line 183 is set to 0, then the output value from output 154 of adder 164 simply passed through adder 165 to input 156 of adder 166. Finally, output terminal 199, connected to the output of last modulo-2 adder 167 in the adder series, outputs the value of the masked PNS.
Microcontroller 200 includes a processor 201, clock 202, and memory 203 interconnected by a bus 204. A power supply 205 provides power to operate the processor, memory and clock. The clock provides timing signals to the processor and memory to synchronize operations. The memory of the microcontroller contains a data module 206 containing the initial values for registers 101-108 and program module 207 to control the processor to transfer those initial values through control lines 135 to those registers at initialization. The memory also includes program module 208 to calculate the mask values in a manner well known in the art, so as to synchronize the respective values provided by the masked PNS of a transmitter and receiver.
In known masked Galois LFSRs, after the LFSR is initialized the mask of the LFSR outputs an invalid sequence until bits of the correct PNS are loaded into all the registers of the mask. This is especially a problem in systems which initialize the LFSR to different initial values during synchronization between transmitters and receivers.
The above citations are hereby incorporated herein in whole by reference.
In the invention of applicants, a masked Galois linear feedback shift register (LFSR) is able to output a correct masked pseudo-noise sequence (PNS) immediately upon initialization of the LFSR. This is accomplished by interconnecting mask switches of a Galois LFSR through a combinatorial network with the registers of the LFSR. The invention also provides a simple method for determining how to interconnect the switches with the registers depending only on the LFSR which in turn depends on the specific primitive polynomial that the Galois LFSR implements.
Those skilled in the art will understand the invention and additional objects and advantages of the invention by studying the description of preferred embodiments below with reference to the following drawings that illustrate the features of the appended claims: